Surface versus deeper level of adaptation

In this review, it was clear that most of the studies used surface modifications, the main focus being translation, with only a few studies also implementing deeper level adaptations. Various steps and measures were undertaken to ensure verbal and written language used in the screening process was clear, understandable, culturally appropriate and syntonic to the individual culture. However, the authors concurred with Soto et al. [29], in that little information was offered to enable conclusions to be drawn on how such adaptations were maintained or guided. For example, the majority of studies mainly reported back and forth translations. Other aspects of surface modifications (e.g., metaphors) were less recognisable, and/or reported in the reviewed studies.

Translation is the first step involved in the adaptation of an instrument. It requires careful planning and equal treatment of linguistic, cultural, contextual and scientific information [60, 61]. Yet, despite the significance of this step, some authors failed to report basic details, such as how many translators were involved and what their qualifications were. Recent evidence indicates the need for a minimum of two bilingual translators, with a cultural background and proficiency in both languages, to minimise the risk of linguistic, psychological, cultural and understanding (i.e., theoretical and practical) biases [60, 62]. Some studies failed to include an expert review or a pilot study. Both steps are essential in synthesising the suitability of an instrument for the targeted cultural context or in approving its readiness for use.

On the other hand, deep levels of adaptation were noted in a few studies throughout the following EVF domains: concepts, contents, methods and context, to redress some cultural and comprehension issues. Among them, the most commonly used domain was adaptation of concepts. In this domain, authors reported efforts to re-word some items in their instruments, using more culturally sensitive concepts to screenings, excluding confusing or difficult items, or presenting participants with clarifying examples framed within the investigated cultural values and traditions [13, 48, 50, 51, 53, 54]. This was followed by an adaptation in methodology that required the researcher to change the screening instrument methodology from a parental report checklist to a trained assessor or interview format. This type of adaptation was undertaken to improve the rigorous nature of the instrument [7, 44, 47, 52, 54]. The least considered domains of deeper adaptation were context and contents. Very few studies considered incorporating information on cultural value, such as the level of education, socio-economic status or the geographic and demographic characteristics of the population [41, 42, 44, 48].

Reviewed studies lacked the justification for favouring a surface type of cultural adaptation. This might be due to the absence of available information on autism screening in each contexts. However, the body of research on autism screening is growing, rapidly and globally. Future studies might be able to identify the level of cultural adaptation and the resources required beforehand. The lack of feasibility studies in this area might be another reason researchers were prevented from conducting deep-level cultural adaptation investigation. This may be due to an inability to estimate expected expenses and required resources for this level of adaptation. In addition to the lack of data on the practicalities of implementing autism screening and the acceptability of screening in the targeted population, cost effectiveness analysis and random controlled trial studies, comparing the satisfaction levels of autism screening groups with that of control groups, might be valuable in advancing this area. Lack of investigator knowledge, interest and expertise on cultural adaptations might be another reason for inadequate documentation and justification for adapting a screening instrument.

Additionally, studies in this review lacked details on the particular cultural adaptation framework that was followed, as well as the efforts taken to avoid bias. This issue was also revealed by Baumann et al. [34], who advocated for an unambiguous description of what had been adapted, why it had been adapted, and how it was adapted. Adequate reporting is necessary for future studies, to promote an effective outcome, maintain high fidelity and avoid decrements in screening impact [34]. The literature also provides a number of guidelines to ensure adequate adaptation is achieved at the process [24, 60, 62, 63] and/or content levels [33, 64]. Noting such guidelines and integrating them within the screening process, may reduce discrepancy among results and enable researchers to replicate studies, and investigate differences between instruments within an increasingly diverse population [62].

Aspects of feasibility on screening for autism

With advances in knowledge of autism screening, identifying the feasibility of this programme has become essential in minimising resource waste, in prioritising decisions and in improving the strength of health care organisations [31]. The studies investigated as part of this review, are generally concerned with the practicalities of screening instruments, in terms of their adequacy, time, cost, and training required to deliver effective screening.

In non-English speaking countries, M-CHAT was a popular screening instrument [7, 41–44, 48, 51, 53, 55, 56], albeit with a number of language and cultural adaptations, as discussed earlier. Adaptation suggests an effectiveness in improving instrument properties (sensitivity, specificity and PPV values), reducing false identification and unnecessary burden. However, this raises concerns regarding costs of training staff and allocating follow-up interviews for parents, especially in those countries with limited staff and resources.

Responding to global use of M-CHAT, a new version of this instrument with 20 items, referred to as a Modified Checklist for Autism in Toddlers, Revised with Follow-Up (M-CHAT-R/F) [65], was released. Despite the existence of this version, all studies in this review, with the exception of a recent one [56] adopted the original version of the M-CHAT, with 23 items. Adopting the new version might reduce the challenges of dealing with some difficult items (i.e., 1 and 4) [43, 50, 53] and providing supporting examples to reduce any future misunderstanding and to improve instrument properties. Carakovac et al. [56] reported less positivity and improved results, when compared with previous M-CHAT studies. Using the revised version of the M-CHAT in future studies might improve its practicality.

Previous M-CHAT studies attributed cultural impacts for any inconsistencies between item responses. However, recent evidence revealed additional reasons, such as demographic characteristics. For example, the level of education might reduce a parents’ ability to understand items in the questionnaire [41, 50, 52]. A lack of parental comprehension might also be the result of parents sharing some autistic characteristics with the child, making it difficult to recognise the abnormal signs of autism [50]. This, therefore, could reduce their credibility, as the sole informants for autism screening and might explain the improved result achieved when a trained assessor or follow-up interview was incorporated into the parental self-report in M-CHAT studies. Another potential reason for discrepancy was the problem of reversed coding for certain items in the M-CHAT, such as items 11, 18 and 22 [48]. Seung et al. [48] recommended adapting and using these items with caution. To improve the practicality of adapting the M-CHAT, issues like these should be considered and investigated, to avoid wasting time, effort, and resources.

Reducing the number of items in other screening instruments found in this review (SRS and FYI) might help speed up the screening process, facilitate its integration into a busy clinical setting, promote the cooperation of parents and make it easier for both parents and professionals, with limited experience, to comprehend the questions and complete them with ease. There appears to be a movement towards the development of screening instruments with less items (e.g., 10 items). The studies have identified the most definitive items that would accurately pinpoint a symptom of autism [7, 11, 46] and proposals to increase their use in screening instruments in the future. Incorporating visual aids (i.e., photographs or pictures) and conceptualising the instrument items using the original language, as was the case in Perera et al. [19], may potentially facilitate a parent’s comprehension and reduce cultural as well as adaptation barriers.

It was evident recent screening scales are moving towards quantitative measurements, with items reorganised as Likert scale types. This was established on the assumption autistic traits normally are distributed in the general population, not only in parents, but also in individuals with no previous diagnosis of autism in their families [50]. Despite advancing knowledge in this area and promising results, these abbreviated quantified instruments warrant further validation globally, considering a participant’s characteristics, such as social factors, cognitive level, and medical history [66]. This will enrich our understanding of the factors that might influence the accuracy of the instruments from a global perspective.

Due to limitations in the screening instruments, scholars have not encouraged integrating autism screening within routine practice and warranted further investigation for individual cultures. For the same reasons, expanding the screening programme beyond the study context and for different populations was not favoured by most researchers, as it indicated limited efficacy.

Cost effectiveness on autism screening is an important practicality to consider when introducing a new programme [39]. However, research in the field of autism screening cost is limited. Assessing the costs of screening might provide a comprehensive insight into the eventual financial burden of both direct (e.g., medical expenditures) and indirect, (e.g., special education/training services, lost productivity by family caregivers [66], parental stress and the hassle of following positive screen participants) factors. Future research is recommended, to adequately compare various screening strategies and potentially identify, the most cost-effective methods for each individual study context. Countries vary significantly in their medical facilities and services. Regions with limited capacity for mental health assessment and services should ensure adequate resources, the sufficient coordination of services in the primary sector and early intervention prior to introducing any autism screening [41, 50, 53]. Future research should investigate the resources and cost effectiveness of introducing autism screening processes into clinical settings, as this will inform and direct future policy decisions. It is also noteworthy that coordination between healthcare and specialised services, in terms of follow-up, along with adequate preparation for early intervention, are crucial in enhancing the benefits of early identification of ASD.

Despite concerns regarding cost, increasing professional awareness and training of professionals to screen for autism were found to be useful in the studies reviewed here, in terms of facilitating the screening process and improving the rigorous nature of the instruments. Kondolot et al. [55] also highlighted the benefits of training staff to screen for autism and the fact it might reduce the high positivity that results from screening instruments used by inexperienced staff or parents. However, studies varied in their documentation of the training received and therefore comparisons could not be made regarding the level of training required or the expenses incurred to facilitate effective screening. Training professionals to recognise early signs of autism is recommended in the clinical guidelines [67], as without standardised training, vital signs and differences in screening results may occur.

Ignored areas in cultural adaptation and feasibilities model

Ultimately, investigators failed to capture some common areas in both models: cultural adaptation (i.e., ‘person and goals’) and feasibility (i.e., ‘acceptability, and demands’). In most studies, participants’ interests, views, perceptions, understanding and agreement, in relation to autism screening goals and implementations, were not discussed or documented much. These areas are essential to bring acceptance to the adapted programme, reducing ethnic and racial discrepancies between investigators and participants, promoting cooperation, increasing demands for autism screening, as well as producing a flexible screening programme, framed within the values, customs and traditions of the targeted population [33]. Future studies should consider both domains when adapting screening programmes to accommodate cultural discrepancies, raising investigator credibility and improving participant and investigator relationships towards an effective outcome.

Strengths and limitations

This review is the first to consider exploring content level of cultural adaptation and the feasibility of screening for autism in non-English-speaking countries, and therein lies its strength. It may serve as a baseline for future practitioners considering adapting an autism screening process for these populations. In terms of rigour, all stages of the process (data selection, extractions and quality validation) were cross-checked by two individuals.

Nonetheless, this review had a number of limitations. No studies written from the non-English speaking literature were included, due to limited resources for translation. The small number of identified studies represents only 13 cultures and thus has limited efficacy globally. However, important and relevant investigated aspects do emerge from the reviewed studies that may guide future work.